The attached patch calculates BPM directly from MIDI clock intervals. When feeding it from Logic over IAC the fluctuations are about +/- 0.5 BPM. You may compare this with your method.

PS. Of course the fluctuations can be reduced by calculating the average of multiple intervals. But this would also reduce the reaction time on fast tempo changes.

Thanks broc - it's essentially the same thing as my patch, except I believe sync~ does what you do with Max objects internally. I thought it (sync~) might be faster, since it operates at audio rate versus scheduler rate.... But basically I get very similar results with each approach...

Yes, I think the BPM jitter directly reflects MIDI jitter. It looks irritating but in general has no audible effect.

Thanks - for my application (a step sequencer), it does have some ramifications though, as any imprecision is additive; if one beat is delayed, then all subsequent beats are delayed...Unless of course some beats are a bit early in coming too. Then I suppose it would all even out in the end.

Yes, it will even out since the jitter consists of +/- deviations. For example, I've calculated the average from 48 clock intervals at 120 BPM and got an exact value of 120.000.

Good to know, thanks.

hi I was wondering, with this patch, how would you modify it so that you get a bang on each 'beat', so when the bpm is 120, it gives 120 bangs a minute? I want to use it to trigger a video effect based on the bpm from an external MIDI clock.

MIDI clock delivers 24 clock events per beat. So a 'beat clock' can simply be implemented with a counter (patch below). You may add a counter reset to 0 if you want stop and restart exactly on the beat.

thanks!